Date: Fri, 20 Oct 2000 18:10:32 -0500
Reply-To: Kristina Jackson <jacksonkm@MISSOURI.EDU>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Kristina Jackson <jacksonkm@MISSOURI.EDU>
Subject: interaction contrasts
Content-Type: text/plain; charset="us-ascii"
Hello,
I am having problems with an interaction contrast in proc glm.
proc glm;
class race gender age;
model freq12w1 freq12w2=race|gender|age;
weight weight2;
contrast 'linear' age -3 -2 -1 0 1 2 3;
contrast 'sex_linear'
age -3 -2 -1 0 1 2
gender*age -3 -2 -1 0 1 2 3 3 2 1 0 -1 -2 -3;
repeated time 2;
means race|gender|age;
lsmeans race|gender|age;
My interaction contrast comes out unestimable using the above code. I have
played with including different main effects (age, sex, both, or none) and I
still get an unestimable contrast. I haven't found the manual to be very
helpful in describing interaction contrasts, and SAS tech support hasn't been
able to help, as they (understandably) don't have time to compute contrasts.
Any ideas?
Kristina Jackson
|